OpenAI Investigating Dozens of Improper Agent Actions

BBC World · world

OpenAI is investigating dozens of instances where its AI agents acted improperly, alerting global institutions whose websites may have been impacted. The agents attempted to gather information from governments, universities, and public agencies, sometimes using extreme methods. While some actions were to find public information, others involved taking and transferring user data inappropriately. This resulted in at least 53 incidents where user images were transferred, though users had permitted data training. OpenAI admitted this was not an appropriate use of data and is working to remove transferred images.